Human Hepatocyte Growth Factor (HGF) is a multifunctional heterodimeric polypeptide produced by mesenchymal cells. HGF has been shown to stimulate angiogenesis, morphogenesis and motogenesis, as well as the growth and scattering of various cell types (Bussolino et al., J. Cell. Biol. 119: 629, 1992; Zarnegar and Michalopoulos, J. Cell. Biol. 129:1177, 1995; Matsumoto et al., Ciba. Found. Symp. 212:198, 1997; Birchmeier and Gherardi, Trends Cell. Biol. 8:404, 1998; Xin et al. Am. J. Pathol. 158:1111, 2001). The pleiotropic activities of HGF are mediated through its receptor, a transmembrane tyrosine kinase encoded by the proto-oncogene cMet. In addition to regulating a variety of normal cellular functions, HGF and its receptor c-Met have been shown to be involved in the initiation, invasion and metastasis of tumors (Jeffers et al., J. Mol. Med. 74:505, 1996; Comoglio and Trusolino, J. Clin. Invest. 109:857, 2002). HGF/cMet are coexpressed, often over-expressed, on various human solid tumors including tumors derived from lung, colon, rectum, stomach, kidney, ovary, skin, multiple myeloma and thyroid tissue (Prat et al., Int. J. Cancer 49:323, 1991; Chan et al., Oncogene 2:593, 1988; Weidner et al., Am. J. Respir. Cell. Mol. Biol. 8:229, 1993; Derksen et al., Blood 99:1405, 2002). HGF acts as an autocrine (Rong et al., Proc. Natl. Acad. Sci. USA 91:4731, 1994; Koochekpour et al., Cancer Res. 57:5391, 1997) and paracrine growth factor (Weidner et al., Am. J. Respir. Cell. Mol. Biol. 8:229, 1993) and anti-apoptotic regulator (Gao et al., J. Biol. Chem. 276:47257, 2001) for these tumors.
HGF is a 102 kDa protein with sequence and structural similarity to plasminogen and other enzymes of blood coagulation (Nakamura et al., Nature 342:440, 1989; Weidner et al., Am. J. Respir. Cell. Mol. Biol. 8:229, 1993, each of which is incorporated herein by reference) (FIG. 1). Human HGF is synthesized as a 728 amino acid precursor (preproHGF), which undergoes intracellular cleavage to an inactive, single chain form (proHGF) (Nakamura et al., Nature 342:440, 1989; Rosen et al., J. Cell. Biol. 127:1783, 1994). Upon extracellular secretion, proHGF is cleaved to yield the biologically active disulfide-linked heterodimeric molecule composed of an α-subunit and β-subunit (Nakamura et al., Nature 342:440, 1989; Naldini et al., EMBO J. 11:4825, 1992). The α-subunit contains 440 residues (69 kDa with glycosylation), consisting of the N-terminal hairpin domain and four kringle domains. The β-subunit contains 234 residues (34 kDa) and has a serine protease-like domain, which lacks proteolytic activity. Cleavage of HGF is required for receptor activation, but not for receptor binding (Hartmann et al., Proc. Natl. Acad. Sci. USA 89:11574, 1992; Lokker et al., J. Biol. Chem. 268:17145, 1992). HGF contains 4 putative N-glycosylation sites, 1 in the α-subunit and 3 in the β-subunit. HGF has 2 unique cell specific binding sites: a high affinity (Kd=2×10−10 M) binding site for the cMet receptor and a low affinity (Kd=10−9 M) binding site for heparin sulfate proteoglycans (HSPG), which are present on the cell surface and extracellular matrix (Naldini et al., Oncogene 6:501, 1991; Bardelli et al., J. Biotechnol. 37:109, 1994; Sakata et al., J. Biol. Chem., 272:9457, 1997). NK2 (a protein encompassing the N-terminus and first two kringle domains of the α-subunit) is sufficient for binding to cMet and activation of the signal cascade for motility, however the full length protein is required for the mitogenic response (Weidner et al., Am. J. Respir. Cell. Mol. Biol. 8:229, 1993). HSPG binds to HGF by interacting with the N terminus of HGF (Aoyama, et al., Biochem. 36:10286, 1997; Sakata, et al., J. Biol. Chem. 272:9457, 1997). Postulated roles for the HSPG-HGF interaction include the enhancement of HGF bioavailability, biological activity and oligomerization (Bardelli, et al., J. Biotechnol. 37:109, 1994; Zioncheck et al., J. Biol. Chem. 270:16871, 1995).
cMet is a member of the class IV protein tyrosine kinase receptor family. The full length cMet gene was cloned and identified as the cMet proto-oncogene (Cooper et al., Nature 311:29, 1984; Park et al., Proc. Natl. Acad. Sci. USA 84:6379, 1987). The cMet receptor is initially synthesized as a single chain, partially glycosylated precursor, p170(MET) (FIG. 1) (Park et al., Proc. Natl. Acad. Sci. USA 84:6379, 1987; Giordano et al., Nature 339:155, 1989; Giordano et al., Oncogene 4:1383, 1989; Bardelli et al., J. Biotechnol. 37:109, 1994). Upon further glycosylation, the protein is proteolytically cleaved into a heterodimeric 190 kDa mature protein (1385 amino acids), consisting of the 50 kDa α-subunit (residues 1-307) and the 145 kDa β-subunit. The cytoplasmic tyrosine kinase domain of the β-subunit is involved in signal transduction.
Several different approaches have been investigated to obtain an antagonistic molecule of the HGF/cMet interaction: truncated HGF proteins such as NK1 (N terminal domain plus kringle domain 1; Lokker et al., J. Biol. Chem. 268:17145, 1993), NK2 (N terminal domain plus kringle domains 1 and 2; Chan et al., Science 254:1382, 1991) and NK4 (N-terminal domain plus four kringle domains; Kuba et al., Cancer Res. 60:6737, 2000), anti-cMet mAbs (Dodge, Master's Thesis, San Francisco State University, 1998) and anti-HGF mAbs (Cao et al., Proc. Natl. Acad. Sci. USA 98:7443, 2001, which is incorporated herein by reference).
NK1 and NK2 can compete effectively with the binding of HGF to its receptor, but have been shown to have partial agonistic activities in vitro (Cioce et al., J. Biol. Chem. 271:13110, 1996; Schwall et al., J. Cell Biol. 133:709, 1996), rather than purely antagonist activities as desired. More recently, Kuba et al., Cancer Res. 60:6737, 2000, demonstrated that NK4 could partially inhibit the primary growth (FIG. 2) and metastasis of murine lung tumor LLC in a nude mouse model by continuous infusion of NK4. The fact that NK4 had to administered continuously to obtain a partial growth inhibition of primary tumors indicates a potentially short half-life of the NK4 molecule and/or lack of potency. Compared to NK4, the approach of using antibodies will benefit from their favorable pharmacokinetics and the possibility of obtaining antibodies with much higher potency.
As another approach, Dodge (Master's Thesis, San Francisco State University, 1998) generated antagonistic anti-cMet monoclonal antibodies (mAbs). One mAb, 5D5, exhibited strong antagonistic activity in ELISA, but induced a proliferative response of cMet-expressing BAF-3 cells, presumably due to dimerization of the membrane receptors. Prat et al., J. Cell Sci. 111:237, 1998, also reported such agonistic activities of anti-cMet mAbs. Zaccolo et al., Eur. J. Immunol 27:618, 1997, used phage display methods do develop human Fab fragments against mouse and human hepatocyte growth factor. These Fab fragments had no effect on the activity of HGF when used alone. When one of the anti-human HGF Fab fragments was combined with an antibody that bound to the Fab fragment itself, it actually enhanced the activity of HGF in a biological assay.
Cao et al., Proc. Natl. Acad. Sci. USA 98:7443, 2001, demonstrated that the administration of a cocktail of three anti-HGF mAbs, which were selected based upon their ability to inhibit the scattering activity of HGF in vitro, were able to inhibit the growth of human tumors in the xenograft nude mouse model (FIG. 3). They postulated that three mAbs recognizing three different binding sites on HGF were required to inhibit the bioactivities of HGF in vivo: two mAbs inhibited the binding of HGF to cMet and one mAb inhibited the binding of HGF to heparin. However, it is impractical for commercial and regulatory reasons to develop a drug combining three novel mAbs, e.g., because some clinical activity of each antibody would need to be demonstrated independently.
Thus, there is a need for a single monoclonal antibody that blocks biological activity of HGF in vitro and in vivo. The present invention fulfills this and other needs.